Audio Event Detection with Window-Based Prediction

3. The computing system of claim 2, wherein a user confidence level corresponding to the class window confidence value is presented to a user and is adjusted by the user.

4. The computing system of claim 2, wherein the class window confidence value for predetermined continuous audio events is increased via a sigmoid function.

5. The computing system of claim 1, wherein the plurality of candidate window sizes, the predetermined smoothing function, and the predetermined prediction threshold are optimized.

6. The computing system of claim 1, wherein the run-time audio signal is a raw audio signal.

7. The computing system of claim 1, wherein each segment of the plurality of segments overlaps with the at least one adjacent segment by 90% or more.

8. The computing system of claim 1, wherein each class of audio event is a defined audio event that is included in an audio track of a video, wherein the defined audio event is editable by a curator of the video using an editor program on the video or a master file for the video.

9. The computing system of claim 1, wherein an individual class of audio event can be characterized by audio features that are shorter duration than a length of the segment.

10. The computing system of claim 1, wherein the predetermined smoothing function is selected from a group consisting of mean, average, max, and identity.

14. The computerized method of claim 13, wherein the semi-synthetic audio signals are randomly or pseudo-randomly inserted into different locations in the raw audio signal.

15. The computerized method of claim 11, wherein each segment of the plurality of segments overlaps with the at least one adjacent segment by 90% or more.

16. The computerized method of claim 11, wherein each class of audio event is a defined audio event that is included in an audio track of a video, wherein the defined audio event is editable by a curator of the video using an editor program on the video or a master file for the video.

19. The one or more storage devices of claim 18, wherein a user confidence level corresponding to the class window confidence value is presented to a user and is adjusted by the user.

20. The one or more storage devices of claim 17, wherein each class of audio event is a defined audio event that is included in an audio track of a video, wherein the defined audio event is editable by a curator of the video using an editor program on the video or a master file for the video.